ExplainabilityTrustworthinessSafetycs.CV
X-Beat: An Explainable Framework for ECG Image Classification
Mohammad Sadman Tahsin, Haitham Y. Adarbah, Afzel Noore
September 18, 2026
Mohammad Sadman Tahsin, Haitham Y. Adarbah, Afzel Noore
Explainability is critical for clinical deployment: The paper demonstrates that high accuracy alone is insufficient for healthcare AI—X-Beat shows that models must also provide visual explanations (using Grad-CAM) of why they make predictions so clinicians can verify that decisions are based on clinically meaningful evidence rather than spurious patterns.
Reliability validation goes beyond accuracy metrics: Beyond reporting 91.94% accuracy, the framework evaluates explanation stability, regional sensitivity, and confidence-based reliability assessment, suggesting that responsible AI in healthcare requires multiple layers of validation to ensure models are trustworthy and robust.
Transfer learning can be effective for medical imaging: The study found that using pre-trained models (ResNet-50, EfficientNet-B0, DenseNet-121, MobileNetV3-Large) trained on general image datasets performed well on ECG classification, indicating that responsible AI development doesn't always require starting from scratch—existing validated approaches can be adapted responsibly.
Model transparency enables clinical verification: By incorporating Grad-CAM visual explanations, the framework allows cardiologists to see which parts of the ECG waveform the model focuses on, creating accountability and enabling clinicians to catch cases where the model might be making correct predictions for wrong reasons.
Systematic benchmarking supports responsible deployment: The unified evaluation framework across multiple models and metrics provides a standardized approach for assessing trustworthiness, which could help healthcare institutions make informed decisions about which AI systems are safe to deploy in clinical settings.
Accurate automated interpretation of electrocardio- grams (ECGs) is essential for early detection of cardiac condi- tions such as myocardial infarction and rhythm abnormalities. However, many high-performing deep learning models remain difficult to deploy in clinical settings due to limited transparency and lack of reliability validation. In this work, we present X- Beat, an explainable and reliability-aware benchmark framework for ECG image classification designed to support trustworthy AI systems in healthcare. The proposed framework combines transfer learning with post-hoc explainability and systematic reliability evaluation across four cardiac classes: Abnormal Heartbeat, History of Myocardial Infarction, Myocardial In- farction, and Normal Heartbeat. Multiple ImageNet-pretrained CNN backbones, including EfficientNet-B0, ResNet-50, DenseNet- 121, and MobileNetV3-Large, are evaluated under a unified training protocol. Beyond standard performance metrics, we incorporate Grad-CAM-based visual explanations together with additional analyses, including explanation stability, regional sen- sitivity, and confidence-based reliability assessment, to examine whether model predictions are supported by clinically meaningful evidence. Experimental results show that ResNet-50 achieves the best performance, reaching 91.94% accuracy and a macro F1- score of 0.9098, with strong class separability (AUC up to 0.995). Explanation analyses indicate that the model primarily focuses on waveform-relevant regions, while reliability evaluation suggests that most incorrect predictions occur with lower confidence. Overall, this work provides a structured and reproducible bench- mark for evaluating both predictive performance and explanation reliability in ECG image classification, contributing toward the development of trustworthy and interpretable AI components for clinical decision support systems.
Accepted to the Workshop on Data Quality Aware, High-Performance, and Trustworthy AI Systems for Healthcare at IEEE/ACM CHASE 2026
Citation data from Semantic Scholar